Incorporating global information into local nonlinear controllers

Abstract

For a particular equilibrium point, the local performance is determined by the partial derivatives of the control law evaluated at the equilibrium point. For a linear controller, the derivatives are equal to the state-feedback gains and the gains on the external inputs. These gains can be changed to vary the local performance of the system. An extended-linear controller links together the desired local controllers of various equilibrium points producing a nonlinear controller with the desired characteristics in a neighborhood of the equilibrium curve.

Global performance is the behavior of the system away from the equilibrium curve. Although the extended-linear controller has good local performance, the global performance might be poor or even unstable. This thesis uses cubic spline techniques to investigate the coupling of global information into local controllers without affecting the local performance. Although “stand-alone” interpolative spline structures do not give the desired local performance, global information can be splined into linear and extended-linear controllers to provide both good global and good local performance.